Published on Web 09/22/2005
Catalysis of Enantioselective [2+1]-Cycloaddition Reactions of Ethyl Diazoacetate and Terminal Acetylenes Using Mixed-Ligand Complexes of the Series Rh2(RCO2)n (L*4-n). Stereochemical Heuristics for Ligand Exchange and Catalyst Synthesis Yan Lou, Travis P. Remarchuk, and E. J. Corey* Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts 02138 Received April 7, 2005; E-mail:
[email protected] Abstract: This paper describes the synthesis of mixed Rh2(II) complexes containing bridging acetate and R,R-diphenyl-N-triflylimidazolidinone (DPTI) ligands (1, 2, and 9-19), and their function as enantioselective catalysts for the conversion of ethyl diazoacetate and terminal acetylenes to chiral cyclopropenes. Of these catalysts, 1 and 10 functioned with the highest enantioselectivity, in accord with a mechanistic model in which one of the ligand bridges is broken in the intermediate Rh-carbene complex. The synthetic results allow conclusions with regard to kinetically and thermodynamically favored pathways for the synthesis of mixed acetate-DPTI complexes. A new C2-symmetric complex having only two anti-DTBTI bridges (23) is shown to be a highly effective chiral catalyst, as expected from the model.
The use of Rh2(II) salts, e.g., Rh2(OAc)4, as catalysts for C-C bond formation in [2+1]-cycloaddition reactions of olefins (or acetylenes) and R-diazo carbonyl compounds or in ring-forming C-H insertion reactions of the latter represents an important synthetic tool,1 made even more powerful by the development of highly enantioselective versions using chiral Rh(II) catalysts. The most effective of these chiral catalysts thus far have been Rh2-bridged dimers having four identical chiral bridging ligands, especially the ligands of McKervey/Davies (N-arylsulfonylproline),1i,2 Doyle (chiral 2-oxopyrrolidines),1a,b,3 and Hashimoto/ Ikegami (N-phthaloyl-tert-butylglycine).1e,4 The rate-limiting step for these catalytic reactions is the reaction of the R-diazo carbonyl with the Rh(II) catalyst, forming N2 and a Rh(II) carbenoid complex.5 The detailed nature of that complex and the subsequent product-forming step, which are both crucial to the understanding of the mechanistic basis for enantioselectivity, (1) For reviews, see: (a) Doyle, M. P.; McKervey, M. A.; Ye, T. Modern Catalytic Methods for Organic Synthesis with Diazo Compounds; John Wiley: New York, 1998. (b) Doyle, M. P.; Ren, T. In Progress in Inorganic Chemistry; Karlin, K. D., Ed.; John Wiley: New York, 2001; pp 113168. (c) Davies, H. M. L.; Beckwith, R. E. J. Chem. ReV. 2003, 103, 28612908. (d) Forbes, D. C.; McMills, M. C. Curr. Org. Chem. 2001, 5, 10911105. (e) Hashimoto, S. Farumashia 2001, 37, 1095-1097. (f) Doyle, M. P.; Forbes, D. C. Chem. ReV. 1998, 98, 911-935. (g) Doyle, M. P.; Protopopova, M. N. Tetrahedron 1998, 54, 7919-7946. (h) Kitagaki, S.; Hashimoto, S. Yuki Gosei Kagaku Kyokaishi 2001, 59, 1157-1168. (i) Davies, H. M. L.; Antoulinakis, E. G. J. Organomet. Chem. 2001, 617618, 47-55. (j) Davies, H. M. L. Eur. J. Org. Chem. 1999, 2459-2469. (2) Kennedy, M.; McKervey, M. A.; Maguire, A. R.; Roos, G. H. P. J. Chem. Soc., Chem. Commun. 1990, 361-362. (3) (a) Doyle, M. P.; Brandes, B. D.; Kazala, A. P.; Pieters, R. J.; Jarstfer, M. B.; Watkins, L. M.; Eagle, C. T. Tetrahedron Lett. 1990, 31, 6613-6616. (b) Doyle, M. P.; Zhou, Q.-L.; Simonsen, S. H.; Lynch, V. Synlett 1996, 697-698. (c) Doyle, M. P.; Winchester, W. R.; Protopopova, M. N.; Mu¨ller, P.; Bernardinelli, G.; Ene, D.; Motallebi, S. HelV. Chim. Acta 1993, 76, 2227-2235. (4) Hashimoto, S.; Watanabe, N.; Ikegami, S. Tetrahedron Lett. 1990, 31, 5173-5174. 10.1021/ja052254w CCC: $30.25 © 2005 American Chemical Society
have been obscure, although the assumption that the carbenoid complex retains the framework of Rh2L4 has commonly been made for symmetrically bridged catalysts.1,6 This assumption has also been used in the latest computational studies of the product-forming step.7,8 We recently have described a different mechanistic model of the product-forming step with unsymmetrically bridged catalysts which is based on the idea that the Rh-carbenoid complex contains only three of the original ligand bridges of the starting Rh(II) catalyst and that the reaction proceeds by a [2+2]-cycloaddition of the CdC or CtC linkage to Rh-carbenoid π-linkage.9,10 On the basis of this hypothesis, we synthesized the chiral Rh(II) complex 1,10,11 having one acetate and three R,R-diphenyl-N-triflylimidazolidinone (DPTI) ligands, and compared it to the closely related complex with (5) (a) Pirrung, M. C.; Morehead, A. T., Jr. J. Am. Chem. Soc. 1994, 116, 8991-9000. (b) Pirrung, M. C.; Morehead, A. T., Jr. J. Am. Chem. Soc. 1996, 118, 8162-8163. (c) Pirrung, M. C.; Liu, H.; Morehead, A. T., Jr. J. Am. Chem. Soc. 2002, 124, 1014-1023. (d) Alonso, M. E.; Garcı´a, M. del C. Tetrahedron 1989, 45, 69-76. (6) (a) Sheehan, S. M.; Padwa, A.; Snyder, J. P. Tetrahedron Lett. 1998, 39, 949-952. (b) Snyder, J. P.; Padwa, A.; Stengel, T.; Arduengo, A. J., III; Jockisch, A.; Kim, H.-J. J. Am. Chem. Soc. 2001, 123, 11318-11319. (7) (a) Nakamura, E.; Yoshikai, N.; Yamanaka, M. J. Am. Chem. Soc. 2002, 124, 7181-7192. (b) Yoshikai, N.; Nakamura, E. AdV. Synth. Catal. 2003, 345, 1159-1171. (8) Nowlan, D. T., III; Gregg, T. M.; Davies, H. M. L.; Singleton, D. A. J. Am. Chem. Soc. 2003, 125, 15902-15911. These workers have also measured 12C/13C kinetic isotope effects for olefin cyclopropanation. (9) For other studies of the conversion of acetylenes to cyclopropenes, see: (a) Petiniot, N.; Anciaux, A. J.; Noels, A.; Hubert, A. J.; Teyssie, P. Tetrahedron Lett. 1978, 1239-1242. (b) Doyle, M.; Protopopova, M.; Mu¨ller, P.; Ene, D.; Shapiro, E. J. Am. Chem. Soc. 1994, 116, 8492-8498. (c) Protopopova, M.; Doyle, M.; Mu¨ller, P.; Ene, D. J. Am. Chem. Soc. 1992, 114, 2755-2757. (d) Mu¨ller, P.; Imogaı¨, H. Tetrahedron: Asymmetry 1998, 9, 4419-4428. (10) Lou, Y.; Horikawa, M.; Kloster, R. A.; Hawryluk, N. A.; Corey, E. J. J. Am. Chem. Soc. 2004, 126, 8916-8919. (11) The structure of this catalyst was confirmed by single-crystal X-ray diffraction analysis. J. AM. CHEM. SOC. 2005, 127, 14223-14230
9
14223
Lou et al.
ARTICLES
four DPTI ligands (2)10,11 in the reaction of ethyl diazoacetate with terminal acetylenes. The chiral complex 1 efficiently
solubility of Rh2(t-BuCO2)4 as compared to Rh2(OAc)4. Of great interest was the finding that 5 catalyzed the formation of S-cyclopropenes from ethyl diazoacetate with about the same enantioselectivity as catalyst 1.10 These results are consistent with a major reaction pathway via 6 (pivalate ligand possibly bidentate), very analogous to that proposed for the reaction with catalyst 1. As described previously, there seems to be no logical explanation of the results obtained with catalyst 5 on the basis of a tetra-bridged carbenoid intermediate since the upper right quadrant of 5 is sterically the most accessible to the acetylene and is also effectively achiral.
catalyzed the [2+1]-cycloaddition of ethyl diazoacetate to a variety of terminal acetylenes to form chiral (S)cyclopropenes with enantioselectivities ranging from 39:1 to 24:1. These enantioselectivities were higher than those obtained
in parallel experiments using the Rh2(DPTI)4 catalyst 2 (ca. 9:1). The excellent results obtained with catalyst 1 are in accord with the triply bridged structure 3 for the Rh-carbenoid intermediate and reaction with the terminal acetylenic substrate by a pathway via assembly 4, in which the more labile acetate bridge to the
rhodium bearing the carbenoid fragment has been broken. The nonbridging acetate ligand in 3 and 4 is arbitrarily shown as monodentate, but it is likely to be attached to the rear rhodium in a bidentate mode. The [2+2]-cycloaddition step 3 f 4 involves the energetically more stable arrangement of the carbenoid fragment, HCCOOEt, with the bulky COOEt group cis to the Rh-O bond and opposite the bulkier N-CHPh group. The regiochemistry of the cycloaddition step is that expected for the carbenoid carbon, being more electrophilic than rhodium. In 3, the Rh bearing the carbenoid has additional electron density in an orbital that can provide better Rh-carbenoid back-bonding than with the corresponding Rh(II) tetra-bridged structure. Even if the tetra-bridged structure were to predominate over the tribridged structure at equilibrium, the reaction with the acetylene would still proceed via 4 if the equilibration were relatively rapid and the reactivity of 4 were greater than that of the isomeric tetra-bridged intermediate. The results obtained with 1 and 2 led us to prepare and examine the behavior of a catalyst containing only two DPTI ligands, specifically the dipivalate complex 5.10,11 The dipivalate was chosen for 5 instead of the corresponding diacetate because it could be prepared in much better yield due to the greater 14224 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005
There are some important points that should be noted with regard to catalysts 1 and 5 and the tri-bridged carbenoid complexes derived therefrom. With regard to 1, it is reasonable to believe that the attack by ethyl diazoacetate occurs at the rhodium having three oxygen substituents and one nitrogen substituent (front Rh in 1), since that rhodium is sterically more available than the other (rear Rh in 1). It is that selectivity which accounts for the preferential formation of the carbenoid complex 3. With reference to catalyst 5, the two Rh centers are equivalent, so it makes no difference which one attaches to the diazo ester. However, the formation of assembly 6 requires that the Rhpivalate bond which breaks preferentially in the carbenoid complex is that which is trans to an oxygen ligand rather than to a nitrogen ligand. Alternatively, this preference may be summarized by stating that the carbenoid is more stable when there is an oxygen trans to the vacant Rh site than when there is a nitrogen. Regardless of the descriptive mode, the preference is a reasonable one, given that the driving force for bridge cleavage is the tendency for minimization of the formal Rh valance state in the carbenoid complex. The present study was initiated with the following objectives: (1) to probe further the mechanistic details of the productforming step in Rh(II)-catalyzed [2+1]-cycloaddition of ethyl diazoacetate to terminal acetylenes; (2) to test the abovedescribed hypotheses regarding the basis for enantioselection using Rh(II) complexes with chiral N-triflylimidazolones, including DPTI and certain structural analogues; (3) to examine the enantioselectivity of [2+1]-cycloaddition for the entire series of mixed DPTI-carboxylate-containing Rh(II) complexes, i.e., the family Rh2(DPTI)n(RCOO)4-n; (4) to examine the factors controlling ligand-exchange reactions of Rh2(II) complexes to clarify the factors controlling the regio- and stereoselectivity; and (5) to identify useful new Rh2(II) catalysts for enantioselective synthesis. The complex demands of rational design of mixed-ligand Rh2(II) complexes have not previously been met with unsymmetrical ligands such as DPTI. Further, very little is known about the rational planning of syntheses of mixed Rh2(II)-carboxylate γ-lactam complexes.
[2+1]-Cycloadditions of Ethyl Diazoacetate and Acetylenes
Synthesis of Mixed Complexes Rh2(OAc)n(OCOCF3)4-n. We carried out initial studies on the synthesis of mixed-ligand Rh2(II) complexes with acetate and trifluoroacetate ligands as a simple model. The selection of this system was also guided by some important findings of J. L. Bear and colleagues on the reaction of Rh2(OAc)4 with trifluoroacetic acid.12,13 These workers reported rate constants for the successive formation of mono-, di-, tri-, and tetratrifluoroacetates to be in the ratio 1:2: 0.1 and 0.025.12 It was of great interest to us that whereas the rate of the second exchange was somewhat faster than the first, the third and fourth exchanges were progressively much slower. Because of this, these researchers were able to isolate a pure mixed complex of composition Rh2(OAc)2(OCOCF3)2, which they considered to be the stereoisomer in which two identical ligands are trans to one another. Upon reinvestigation of this interesting reaction, we found that exposure of Rh2(OAc)4 to excess CF3CO2H at 23 °C for 1.5-2 h produced this major product which could be isolated in 51% yield after column chromatography on silica gel.14 Recrystallization from 10%
CH3OH in CH2Cl2 afforded monoclinic crystals which were subjected to X-ray crystallographic analysis and shown to be the cis isomer of Rh2(OAc)2(OCOCF3)2 (7)15 rather than the trans isomer.12 A simple explanation for this may be that the acetate bridge which is trans to the relatively electronegative trifluoroacetate in Rh2(OAc)3(OCOCF3) is bound more tightly than the other two and thus is more difficult to protonate (in the intermediate for ligand exchange, presumably having two nonbridging axial CF3CO2 ligands coordinated to the tetrabridged Rh2(OAc)3(OCOCF3)) and to displace. In any event, it is clear that the more electron-attracting CF3CO2 bridge disfavors displacement of the trans-CH3CO2 bridge relative to the two cis-CH3CO2 bridges. These observations suggested a rational plan for the synthesis of trans-Rh2(OAc)2(OCOCF3)2 (8) that proved to be successful. Reaction of Rh2(OCOCF3)4 in CH3CN solution with 2 equiv of n-Bu4N+OAc- at 0 °C afforded, after column chromatography, 75% yield of 8, the structure of which was proven by
ARTICLES
single-crystal X-ray diffraction analysis15,16 (crystallized from CH3CN/C6H6). This preparative route takes advantage of the more facile displacement of the trans-trifluoroacetate bridge in Rh2(OCOCF3)3(OAc) by AcO- and also the obvious driving force for displacement of the more stabilized trifluoroacetate ion by the less stabilized acetate ion. The facile preparation of the cis (7) and trans (8) isomers of Rh2(OAc)2(OCOCF3)2 provided both a heuristic for synthetic planning and two very useful starting materials for the synthesis of mixed-ligand Rh2(II) complexes. Synthesis of Mixed-Ligand Rh2(II) Complexes with DPTI and Acetate. There are 14 possible Rh2(II)-DPTI complexes, including mixed-ligand complexes with acetate, having the formula Rh2(OAc)n(DPTI)4-n. These are systematically displayed in Scheme 1 as a synthetic tree which shows the possible pathways of formation by unidirectional ligand displacement. We have been able to synthesize all but one of these, complex 20. The structures of complexes 11,11 1,10 15,11 19,11 and 210 were determined by single-crystal X-ray diffraction analysis, as indicated in Scheme 1. Structure 10 is the acetate analogue of pivalate 5, the structure of which was also determined by X-ray analysis,10 and gave essentially identical ee values in [2+1]-cycloaddition reactions with terminal acetylenes. The ee values given in Scheme 1 represent those experimentally determined for the various complexes as catalysts for the reaction of ethyl diazoacetate and 1-heptyne in CH2Cl2 at 23 °C under the same standardized conditions. Structural assignments to the remaining complexes in Scheme 1 will be discussed below together with the method of synthesis. A mixture of the three cis isomers of Rh2(OAc)2(DPTI)2, 10, 11, and 12, was synthesized in a logical way from the cis isomer of Rh2(OAc)2(OCOCF3)2 (7) by reaction with the sodium salt of DPTI (prepared by reaction of DPTI in THF with sodium hexamethyldisilazane, NaN(TMS)2) in THF solution at 0 °C for 2 h and at 23 °C for 12 h, as shown in Scheme 2. Purification of the crude reaction product (ratio of 10, 11, and 12 ca. 4:2:1) by chromatography on silica gel (5% CH3CN in CH2Cl2 for elution) afforded the following isolated yields: 10, 43%; 11, 23%; and 12, 11%. The structure of 12 was obvious from the NMR spectrum, which shows nonequivalent DPTI ligands (e.g. two CF3 peaks in the 19F NMR spectrum in contrast to the spectra 10 and 11, which show only a single sharp CF3 peak), and from the fact that the cyclopropene that formed from ethyl diazoacetate and 1-heptyne with 12 as catalyst was completely racemic. A racemic product is to be expected since the Rhcarbenoid intermediate from 12 is expected to be that in which the carbene has attached to the Rh that initially had four oxygen ligands (the sterically most accessible and the most Lewis acidic Rh). Such a carbenoid clearly has a symmetric environment that would not lead to enantioselection in [2+1]-cycloaddition reactions. The structure of 10 follows logically not only by a process of elimination but also from its identical catalytic behavior with the dipivalate analogue 5, to which that structure had been assigned unambiguously by X-ray analysis.10 (12) Bear, J. L.; Kitchens, J.; Willcott, M. R., III. J. Inorg. Nucl. Chem. 1971, 33, 3479-3486. (13) See also: Johnson, S. A.; Hunt, H. R.; Neumann, H. M. Inorg. Chem. 1963, 2, 960-962. (14) Small amounts of Rh2(OCOCF3)3(OAc) and Rh2(OAc)3(OCOCF3) were also isolated from the reaction mixture by chromatography. (15) For details on the determination of structure by single-crystal X-ray diffraction analysis, see Supporting Information. (16) The only other reaction product was Rh2(OAc)3(OCOCF3). J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005 14225
Lou et al.
ARTICLES Scheme 1. Synthetic Tree for Sequential Formation of Rh2(OAc)n(DPTI)4-n Complexesa
a The ee values shown in Scheme 1 refer to those measured for the catalyzed addition of ethyl diazoacetate to 1-heptyne (CH Cl at 23 °C). b At 40 °C 2 2 in CH2Cl2.
The cis-anti complexes 10 and 11 could also be prepared in one step from Rh2(OAc)4 by heating with 2 equiv of DPTI-H in chlorobenzene-acetic acid at reflux for 36 h. After chromatography of the crude reaction product, 10 was obtained in 25% isolated yield and 11 was obtained in 31% isolated yield. The remaining rhodium is accounted for as Rh2(OAc)4 and Rh2(OAc)3(DPTI) under these conditions. The acetic acid cosolvent minimizes the amount of Rh2(OAc)(DPTI)3 formed under these conditions. Reaction of trans-Rh2(OAc)2(OCOCF3)2 (8) with Na+DPTIin THF at 23 °C gave cleanly trans-syn-Rh2(OAc)2(DPTI)2 (14).
Scheme 2
14226 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005
[2+1]-Cycloadditions of Ethyl Diazoacetate and Acetylenes
The preferential formation of the trans-syn isomer 14 over the trans-anti isomer 13, which is thermodynamically more stable than 14, is especially interesting (see below). In the conversion 8 f 14, the anion of DPTI preferentially displaces trifluoroacetate rather than acetate because the former is a better leaving group. Since trans-anti-Rh2(OAc)2(DPTI)2 (13) was not available from 8, an alternative route was devised for its synthesis starting from the readily available 1. Upon heating of 1 with 10 equiv of n-Bu4N+OAc- and 1 equiv of Rh2(OAc)4 in Et2O at reflux for 12 h, trans-anti-Rh2(OAc)2(DPTI)2 could be obtained in 20% isolated yield after chromatographic separation on silica gel.
The remaining material from the reaction was the starting complex Rh2(OAc)(DPTI)3 (1). Subsequent to the development of this first preparation of 13, it was discovered that the same compound could be made by isomerization of the less stable 14. Thus, upon heating of 14 in chlorobenzene solution at reflux for 10 h, it was transformed into an 82:18 mixture of 13 and 14, respectively. The structures of 13 and 14 followed from 1H and 19F NMR studies in the presence of pyridine. Complex 13, in which the two DPTI ligands are in the transanti relationship, coordinates with 1 equiv of pyridine to generate a mono-pyridine adduct that exhibits four distinct methine signals in the 1H NMR spectrum. With excess pyridine a symmetrical bis-pyridine adduct is formed showing two methine signals. In contrast, complex 14, in which the two DPTI ligands are in a trans-syn relationship, coordinates with 1 equiv of pyridine selectively to form a single mono-pyridine adduct that shows two methine signals, as expected for the complex of pyridine at the Rh having four oxygen ligands. The bis-pyridine complex of 14 also shows two methine proton signals, as expected for that structure. It is noteworthy that when 14 was employed as catalyst in the [2+1]-cycloaddition of ethyl diazoacetate to 1-heptyne, the cyclopropene carboxylic ester was produced with only 5% ee, as expected for trans-syn-Rh2(OAc)2(DPTI)2. The remaining rhodium-DPTI complexes that appear in Scheme 1 were readily synthesized. Reaction of either complex 10 or 11 with Na+DPTI- in THF at 50 °C afforded 16, allowing assignment of its structure. Complex 17 was similarly prepared from 12 and Na+DPTI- in THF, which allowed assignment of the structure 17. Complexes 18 and 19 were formed along with 2 by heating Rh2(OAc)4 with DPTI in chlorobenzene at reflux in a Soxhlet apparatus containing a mixture of CaH2 and Celite 545 to remove HOAc from the reaction mixture. Chromatography of the mixture furnished 2 as major product, lesser amounts of 18, and an even smaller amount of 19. The structure of 18 followed clearly from the 1H and 19F NMR spectra, which showed two sets of peaks due to the two pairs of identical ligands (i.e., those trans to each other). Structure 20, the only remaining possibility, would show only a single set of peaks
ARTICLES Table 1. Anionic Ligand Displacement of Acetate by DPTI Anion in THF: Observed Ratio of Product to Remaining Starting Material (9) as a Function of Time t, min
9
10
11
12
13
14
10 30 60 120 180a 600b
1 1 1 1 1 1
0 0.02 0.04 0.09 0.13 0.31
0 0.03 0.05 0.12 0.19 0.53
0 0.02 0.03 0.05 0.06 0.10
0 0 0 0 0.03 0.20
0 0 0 0 0.15 0.49
a
1 (0.36) and 16 (0.03) also formed. b 1 (0.61) and 16 (0.16) also formed.
from the four equivalent ligands. We have never observed 20 as a reaction product, in all likelihood because that structure is destabilized by strong steric repulsions between the substituents on the four bridges and also destabilized electronically (see below). Enantioselectivity of the Reaction of Ethyl Diazoacetate with 1-Heptyne as a Function of Catalyst Structure with Complexes 1, 2, and 9-19. It is apparent from the data shown in Scheme 1 that the highest enantioselectivities in the catalytic test reaction of ethyl diazoacetate with 1-heptyne were observed with catalysts 1 (95% ee) and 10 (91% ee), as expected for a pathway via pre-transition-state complexes 4 and 6 (acetate instead of pivalate), respectively, and as discussed in the introductory section and in an earlier publication.10 The lower enantioselectivities found for the catalyzed ethyl diazoacetateheptyne reactions using the rhodium complexes 11-19 and 2 (summarized in Scheme 1) are also consistent with the proposed pathway. It is interesting that of all the Rh2(II) catalysts shown in Scheme 1, the least reactive toward ethyl diazoacetate is the cis-(2,2)-18, in which every Rh-O bond is trans to an Rh-N bond. In the case of trans-anti-Rh2(OAc)2(DPTI)2 (13) as catalyst, the 82:18 selectivity (64% ee) for formation of ethyl (1S)-2-n-amyl-2-cyclopropenylcarboxylate from 1-heptyne and ethyl diazoacetatesour mechanistic modelsrequires selectivity in the cleavage of one acetate bridge in the intermediate carbenoid (the upper acetate bridge of 13 in Scheme 1). This point is discussed further in the last section of this paper. The very low ee’s (0-5%) that result from the use of catalysts 12, 14, and 17 are predictable from the mechanistic model. Preferred Pathways for the Formation of Rh2(OAc)n(DPTI)4-n Complexes by DPTI- Displacement of Acetate. We have also carried out a kinetic investigation of selective formation of mixed acetate-DPTI complexes of Rh(II) using the displacement reaction of acetate by Na+DPTI- in THF solution. The greater stability of AcO- vs DPTI- clearly provides the driving force for this process. The reaction of Rh2(OAc)n(DPTI)4-n with Na+DPTI- (from DPTI and sodium hexamethyldisilazane) in THF at 50-60 °C was monitored by HPLC analysis to determine quantitatively the rate of formation of the new complexes formed by ligand displacement. The complexes 9-19, 1, and 2 could all be distinguished by HPLC analysis using a standard elution protocol.17 Some of the salient results of this study are presented in Tables 1 and 2. A more (17) HPLC analysis was performed using a Microsorb MV silica column, 23 °C, detection at 254 nm, 1 mL/min flow rate, with the following solvent mixtures: 20/78/2 EtOAc-hexanes-CH3CN (30 min); 80/18/2 EtOAchexanes-CH3CN (30 min); then 20/78/2 EtOAc-hexanes-CH3CN (10 min). The retention times of the isomers in Scheme 1 were determined to be as follows: 9, 51 min; 10, 39.8 min; 11, 40.5 min; 12, 41 min; 13, 22 min; 14, 25 min; 1, 15.2 min; 15, 13.7 min; 16, 10.3 min; 17, 30.3 min; 18, 10.0 min; 19, 6.1 min; 2, 10.8 min; (R,R)-DPTI, 7.0 min. J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005 14227
Lou et al.
ARTICLES Table 2. Anionic Ligand Displacement of Acetate by DPTI Anion in THF: Observed Ratio of Product to Remaining Starting Material (10, 11, or 12) as a Function of Time a. From 10
Scheme 4. Other Pathways for Anionic Ligand Displacement of Acetate by DPTI Anion in THF
b. From 11
t, min
10
1
16
11
16
2
60 180 1440
1 1 1
0 0 0.76
0.24 0.36 11.4
1 1 1
0 0.06 0.39
0 0 0.41
t, min
10
11
1
15
16
17
30a
0 0.01 0.02
1 1 1
0.08 0.09 0.12
0.47 0.46 0.49
3.2 3.4 3.5
0.88 0.97 0.95
c. From 12
60a 180a a
No 12 remaining.
Scheme 3. Kinetically Preferred Pathways for Anionic Ligand Displacement of Acetate by DPTI Anion in THF
extensive summary of the data appears in the Supporting Information. From perusal of the information in Tables 1 and 2, it is evident that certain regio- and stereochemical preferences exist for the ligand displacement reactions and that some products are preferred over the other possibilities that are summarized in Scheme 1. These kinetically favored reaction pathways are shown in Scheme 3. Starting from Rh2(OAc)3(DPTI), the Rh2(OAc)2(DPTI)2 isomers 10, 11, and 12, all of which have a cis arrangement of ligands (two bridges are cis if they are at 90° angles to one another), are favored over the trans isomers 13 and 14. At long reaction times 13 and 14 appear, evidently as secondary products formed by isomerization of 10, 11, and/or 12. The Rh2(OAc)2(DPTI)2 isomers 10 and 11 are both converted preferentially to 16 by acetate displacement. Complex 16 is then preferentially converted to 2 by displacement of acetate by DPTI-. Our analyses also revealed 14228 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005
that 16 can undergo an interesting isomerization to 1 and that 1 is converted preferentially to 2. This isomerization of 16 to 1 indicates that it is possible for DPTI- to displace itself and to do so in a way that effects a syn/anti stereomutation. The behavior of the cis-syn-Rh2(OAc)2(DPTI)2 complex 12 is also interesting. Upon heating with Na+DPTI- in THF, it is rapidly isomerized to the clearly more stable P-cis-anti complex 11, another example of stereomutation by DPTI- at a rate that is fast relative to displacement of AcO-. Our kinetic studies have also revealed a number of interconversions that result from slower pathways for nucleophilic displacement of acetate by DPTI- in Rh(II) complexes. These slower pathways are summarized in Scheme 4. In addition to the isomerization process that converts 12 to 11, 12 undergoes displacement to form 15 and 17. The trans-syn complex 14 also is subject to isomerization to form the trans-anti complex 13 that then goes on to 15 by acetate displacement. It is of considerable interest that trans-syn-Rh2(OAc)2(DPTI)2 (14) is less stable than the trans-anti isomer 13 because this difference in stability must be associated with intrinsically less favored bonding for DPTI ligands that are syn to one another as compared to anti since steric destabilization cannot be a factor here.18 Our failure to synthesize complex 20 may be due to the occurrence in 20 of both electronic destabilization of the type operating in 14 and steric repulsion. General Aspects of Rh2(II) Ligand-Exchange Reactions. The results described above on the displacement of acetate in Rh2(II) complexes by Na+DPTI- in THF at 50-60 °C point to some significant trends with regard to preferred pathways. First, the formation of cis-anti arrangements of DPTI ligands appears to be favored. Second, the displacements are kinetically controlled at low conversions. Third, Na+DPTI- can also cause isomerization of unstable to more stable structures in the Rh2(OAc)2(DPTI)2 series, for example the conversion of 12 to 11, or 14 to 13. The conventional method of synthesis of Rh2(II) complexes with chiral ligands involves the thermal reaction of an achiral rhodium carboxylate, usually Rh2(OAc)4, with the chiral ligand (18) It is possible that the trans-anti arrangement of DPTI ligands is energetically more favorable than the trans-syn alternative in Rh2(II) complexes because the former allows the two Rh centers to have the same electron density and effective charge.
[2+1]-Cycloadditions of Ethyl Diazoacetate and Acetylenes
in refluxing solvent (e.g., C6H6 or C7H8). We have used this method extensively to synthesize Rh2(DPTI)4 and Rh2(OAc)(DPTI)3 complexes, generally using C6H5Cl as solvent at reflux with external CaH2-Celite 545 for continuous removal of HOAc. Careful study of this thermal process by HPLC analysis of reaction mixtures as a function of time has demonstrated that this method generally leads to near-equilibrium mixtures of products (see Supporting Information). Studies on trans-anti-Rh2(OAc)2(DTBTI)2. As indicated in Scheme 1, trans-anti-Rh2(OAc)2(DPTI)2 (13) catalyzes the reaction of ethyl diazoacetate and 1-heptyne to form ethyl (1S)2-n-amyl-2-cyclopropenylcarboxylate of 64% ee (82:18 selectivity). Because this enantioselection can be explained by a preference for pre-transition-state assembly 21 over the alternative 22, we were interested in testing whether the difference in stability of 21 and 22 (or their bidentate acetate equivalents) could be magnified by replacing the phenyl groups in catalyst 13 by the bulkier tert-butyl group, as shown in structure 23.
ARTICLES Scheme 5
Table 3. Cyclopropenation of Ethyl Diazoacetate and Terminal Alkynes Catalyzed by 23
entry
R
T, °C
yield, %
ee, %a
1 2 3 4
CH3(CH2)4 CH3(CH2)4 t-Bu MeOCH2
23 0 0 0
87 84 81 78
89 91 90 93
a
That replacement would be expected to disfavor a pathway via the tert-butyl analogue of 22 vs that via the tert-butyl analogue of 21 for two reasons: (1) the steric repulsion between the acetylenic R group and tert-butyl in the analogue of 22 should be considerably greater than for phenyl in 22 and (2) the steric repulsion between the nonbridging acetate ligand in the tertbutyl analogue of 22 would be greater than for phenyl in 22. Consequently, we have synthesized the trans-anti complex 23, which we designate herein as trans-anti-Rh2(OAc)2(DTBTI)2, where DTBTI stands for the ligand (R,R)-4,5-di-tert-butyl-Ntriflylimidazolinone (25). The synthesis of the (R,R)-ligand 25 was carried out from (1R,2R)-1,2-di-tert-butylethylenediamine (24).19 The route of synthesis is outlined in Scheme 5. The reaction of trans-Rh2(OAc)2(OCOCF3)2 with Na+DTBTI- in THF at 23 °C gave a mixture of trans-anti-Rh2(OAc)2(DTBTI)2 (23) and trans-syn-Rh2(OAc)2(DTBTI)2 in a ratio of ca. 1:2.20 Pure 23 was obtained by chromatography of this mixture on silica gel (26% yield, unoptimized).21 (19) Diamine 24 was prepared in this laboratory in 1988 by Dr. Po-Wei Yuen by addition of tert-butylmagnesium chloride and the Schiff base of benzylamine, glyoxal, followed by debenzylation with H2 and Pd-C catalyst, and then resolution with R,R-tartaric acid, a method subsequently developed also by Roland et al. (Roland, S.; Mangeney, P.; Alexakis, A. Synthesis 1999, 228-230). (20) As described above, the corresponding reaction in the DPTI series gave selectively trans-syn-Rh2(OAc)2(DPTI)2 (14).
Enantiomeric excess was determined by GC using a γ-TA column.10
As expected from our mechanistic model, trans-anti-Rh2(OAc)2(DTBTI)2 (23) afforded substantially better enantioselectivity than trans-anti-Rh2(OAc)2(DPTI)2 (13) for the reaction of ethyl diazoacetate with 1-alkynes. Table 3 summarizes the data obtained with catalyst 23 and three different alkynes which show enantioselectivities of approximately 20:1 for each case. We believe that these results provide additional evidence in support of a pre-transition-state model analogous to 21 (for the DPTI series). Conclusions. The major objective of this work was the investigation of the synthesis and catalytic properties of a series of complexes of the type Rh2(OAc)n(ligand)4-n. Both cis- and trans-Rh2(OAc)2(OCOCF3)2 have been synthesized, analyzed to demonstrate structure, and applied as starting materials to the synthesis of chiral Rh2(II) complexes. Thirteen of the complexes Rh2(OAc)n(DPTI)4-n, whose structures are systematically displayed in Scheme 1, have been made and characterized structurally and then applied as catalysts to determine their effectiveness in the synthesis of chiral ethyl 2-n-amyl-2cyclopropenylcarboxylate from 1-heptyne and ethyl diazoacetate. As predicted from the mechanistic model described in the introduction, complexes 1 and 10 excel with regard to enantioselectivity, complexes 12, 14, 17, and 19 are poor, and 2, 11, 13, 15, 16, and 18 are mediocre (64-80% ee). Also as expected from the mechanistic model, trans-anti-Rh2(OAc)2(DTBTI)2 (23) gave much better enantioselectivity (91%) than the DPTI (21) The structures of 23 and trans-syn-Rh2(OAc)2(DTBTI)2 were determined by X-ray diffraction analysis. Interestingly, the latter forms only a monocomplex with either MeOH or pyridine, as shown by X-ray or 1H NMR spectral analysis, respectively. (22) After this paper was submitted for publication, a Communication appeared in which 12,13C kinetic isotope effects were measured for Rh2(OAc)4 and Rh2(OAc)(DPTI)3 as catalysts for the reaction of N2CHCOOEt with 1-pentyne. See: Nowlan, D. T.; Singleton, D. A. J. Am. Chem. Soc. 2005, 127, 6190-6191. These results and theoretical calculations were claimed to support the pathway involving tetrabridged Rh-carbenoid species, contrary to our proposal.10 In our opinion, neither the kinetic isotope effect data nor the theoretical calculations provide an adequate basis for favoring one pathway over the other. J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005 14229
Lou et al.
ARTICLES
analogue 13 (64% ee). Preferred stereochemical pathways were identified for the anionic displacement of acetate by Na+DPTIin the synthesis of the complexes shown in Scheme 1. This work provides a basis for more rational planning of the stereocontrolled synthesis of complexes of the series Rh2(OAc)n(ligand)4-n.22
Supporting Information Available: Synthetic procedures and characterization data for the new Rh2(II) complexes reported herein (PDF); X-ray crystallographic data for compounds 1, 2, 7, 8, 11, 15, 19, 23, and trans-syn-Rh2(OAc)2(DTBTI)2 (CIF). This material is available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We are grateful to Drs. Richard Staples and Robin Kloster for the X-ray crystallographic data.
JA052254W
14230 J. AM. CHEM. SOC.
9
VOL. 127, NO. 41, 2005
